On Wed, Aug 23, 2000 at 10:52:28PM +0200, Stefan Westerfeld wrote:
>    Hi!
> 
> Here is yet-another-patch for making mpeglib resample correct in those cases
> where the samplingrate is something like 32000. The essential part here is
> not the float -> double conversion but this section:
>

I have applied it.

Martin
 
>      // calculate how many input samples we need, then to satisfy the request
> -    float wantWavSamples = ((float)samples+byteMultiplikator)*speed;
> +    // use a larger amount than "really" required, to ensure that samples are
> +    // available for rounding errors and interpolation
> +    double wantWavSamples = (double)samples*speed+8.0;
>  
> Adding 8.0 should be a pretty safe choice for covering up both: float -> int
> rounding errors here and in the actual conversion, and the fact that you need
> more than one sample to do interpolation (currently one more due to linear
> interpolation). The byteMultiplikator seems to have nothing to do with the
> problem here, so I removed it.
> 
> Full patch is attached.
> 
>    Cu... Stefan
> -- 
>   -* Stefan Westerfeld, stefan@space.twc.de (PGP!), Hamburg/Germany
>      KDE Developer, project infos at http://space.twc.de/~stefan/kde *-         

> Index: decoderBaseObject_impl.cpp
> ===================================================================
> RCS file: /home/kde/kdemultimedia/mpeglib_artsplug/decoderBaseObject_impl.cpp,v
> retrieving revision 1.9
> diff -u -r1.9 decoderBaseObject_impl.cpp
> --- decoderBaseObject_impl.cpp	2000/08/17 12:01:52	1.9
> +++ decoderBaseObject_impl.cpp	2000/08/23 20:38:17
> @@ -286,10 +286,12 @@
>      
>      // calculate "how fast" we consume input samples (2.0 means, we need 2
>      // input samples to generate one output sample)
> -    float speed = (float)wav_samplingRate / samplingRateFloat;
> +    double speed = (double)wav_samplingRate / (double)samplingRateFloat;
>      
>      // calculate how many input samples we need, then to satisfy the request
> -    float wantWavSamples = ((float)samples+byteMultiplikator)*speed;
> +    // use a larger amount than "really" required, to ensure that samples are
> +    // available for rounding errors and interpolation
> +    double wantWavSamples = (double)samples*speed+8.0;
>  
>      // convert that into bytes and try to read that many bytes
>      wantBytes=(int) (wantWavSamples*byteMultiplikator);
> @@ -308,7 +310,7 @@
>      //haveSamples=samples;
>      // calculate where we are now (as floating point position) in our
>      // inputsample buffer
> -    flpos += (float)haveSamples * speed;
> +    flpos += (double)haveSamples * speed;
>      
>      // Good - so how many input samples we won't need anymore (for the
>      // next request)? Skip them.
> Index: decoderBaseObject_impl.h
> ===================================================================
> RCS file: /home/kde/kdemultimedia/mpeglib_artsplug/decoderBaseObject_impl.h,v
> retrieving revision 1.4
> diff -u -r1.4 decoderBaseObject_impl.h
> --- decoderBaseObject_impl.h	2000/07/18 19:23:47	1.4
> +++ decoderBaseObject_impl.h	2000/08/23 20:38:18
> @@ -49,7 +49,7 @@
>    DecoderPlugin* decoderPlugin;
>    InputStream* inputStream;
>    ArtsOutputStream* outputStream;
> -  float flpos;
> +  double flpos;
>    float startTime;
>    int instance;
>    int lastAudioBufferSize;

_______________________________________________
Kde-multimedia mailing list
Kde-multimedia@master.kde.org
http://master.kde.org/mailman/listinfo/kde-multimedia